Expt1- Recognition Memory

In this preregistered experiment we examined the effect of testing expectancy on the disfluency effect using a recognition memory test.

Load in packages

Read in raw groilla data

# read in low test expect data exported from gorilla
setwd(here::here('expt1_recog_data', 'Gorilla_data_low'))

data=here::here('expt1_recog_data', 'Gorilla_data_low')  # path to data files

file_list=list.files(data, pattern=".csv") # list of data files
 
# read in all files
datasetlow <-
  do.call("rbind", lapply(file_list, FUN=function(files){
    
    for (i in 1:length(files)){ 
      if(file.exists(files[i])){
        message( "now processing:", files[i])
      }
    }
    fread(files, header=TRUE, sep=",", na.strings = "", fill=TRUE)})) #fread makes reading in files quick
#

library(lubridate)
# clean up data! Select data from after the pre-registation! 


low<-datasetlow %>% 
  janitor::clean_names(.) %>% 
  dplyr::mutate(date=as.Date(utc_date)) %>%
  dplyr::filter(date=="08/06/2020" |date=="09/06/2020" , zone_type=="response_button_text")

#response as character
low$response<-as.character(low$response)

#assign column to denot low test expect
low$testexpect<-"low"

# high test expect
setwd(here::here('expt1_recog_data', 'Gorilla_data_high'))

data=here::here('expt1_recog_data', 'Gorilla_data_high')  # path to data files

file_list=list.files(data, pattern=".csv") # list of data files
 
# read in all files
highdata <-
  do.call("rbind", lapply(file_list, FUN=function(files){
    
    for (i in 1:length(files)){ 
      if(file.exists(files[i])){
        message( "now processing:", files[i])
      }
    }
    fread(files, header=TRUE, sep=",", na.strings = "", fill=TRUE)})) #fread makes reading in files quick
#

library(lubridate)
# a batch of Ss we run before preregistration that should not be included in the analysis
high <-highdata %>% 
  janitor::clean_names(.) %>% 
  dplyr::mutate(date=as.Date(utc_date)) %>%
  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020" |date=="0010/07/2020" | date=="09/06/2020", zone_type=="response_button_text")

#response as character
high$response<-as.character(high$response)

# assign column to denot high test expect
high$testexpect<-"high"

# bind low and high datasets
high_low<-rbind(high, low)

Recognition memory

Calculate d-prime

#response as character
#calculate hit rate and far and compute dprime and other measures
ex4=high_low %>% dplyr::mutate(condition1= dplyr::case_when( 
  condition == "SF" ~ "Sans Forgetica", 
  condition =="normal" ~  "Arial", 
), isold= dplyr::case_when (
  old_new== "old" ~ 1, 
  old_new== "new" ~ 0), 
sayold=dplyr::case_when( 
  response=="old"~ 1, 
  response=="new" ~ 0, 
  ))


#classic SDT
sdt <- ex4 %>% 
  dplyr::mutate(type = "hit",
         type = ifelse(isold==1 & sayold==0, "miss", type),
         type = ifelse(isold==0 & sayold==0, "cr", type),  # Correct rejection
         type = ifelse(isold==0 & sayold==1, "fa", type))  # False alarm
sdt <- sdt %>% 
  dplyr::group_by(participant_private_id, type, condition1, testexpect) %>% 
  dplyr::summarise(count = n()) %>% 
  tidyr::spread(type, count)  # Format data to one row per person

sdt <- sdt %>% 
  dplyr::group_by(participant_private_id, condition1, testexpect)%>%
  dplyr::mutate(hr = hit / (hit+miss),
         fa = fa / (fa+cr)) %>%
  dplyr::mutate(hr=case_when(
    is.na(hr) ~ 0.99,
    TRUE ~ hr), 
    fa=case_when(
      is.na(fa)  ~ 0.01,
    TRUE ~ fa),
     zhr=qnorm(hr), 
     zfa=qnorm(fa), 
    dprime = zhr-zfa) %>%
  ungroup()

Plot

Raincloud plots (Allen et al., 2019) depicting raw data (dots), box plots, and half violin kernel desntiy plots, with mean (red dot). Proportion of “old” responses as a function of Test Expectancy for Experiment 1.

#set up raincloud params
# fig for dprime

highlowaov=sdt  %>% select(participant_private_id, condition1, testexpect, dprime) %>%
  mutate(testexpect=ifelse(testexpect=="low", "Low Test Expectancy", "High Test Expectancy"))

#plot
bold <- element_text(face = "bold", color = "black", size = 14)
sdtmean= highlowaov %>%
  dplyr::group_by(testexpect, condition1) %>%
  dplyr::summarise(mean1=mean(dprime))

## `summarise()` regrouping output by 'testexpect' (override with `.groups` argument)

sdt1=sdt  %>% select(participant_private_id, condition1, testexpect, hr, fa) %>% 
  pivot_longer(hr:fa, names_to="type") %>%
  dplyr::mutate(isold=case_when(type=="hr" ~ "Old", type=="fa" ~ "New"))

sdt1$isold<-factor(sdt1$isold, levels=c("Old", "New"))

sdt1$Condition<-factor(sdt1$condition1, levels=c("Arial", "Sans Forgetica"))


highlowaov=sdt  %>% select(participant_private_id, condition1, testexpect, dprime) %>%
  mutate(testexpect=ifelse(testexpect=="low", "Low Test Expectancy", "High Test Expectancy"))

highlowaov_wide<- highlowaov %>% 
  tidyr::pivot_wider(names_from = "condition1", values_from = "dprime") %>%
  dplyr::mutate(Difference=`Sans Forgetica` - Arial)

highlowwide_mean <- highlowaov_wide %>%
  dplyr::group_by(testexpect) %>%
  dplyr::summarise(mean=mean(Difference))

## `summarise()` ungrouping output (override with `.groups` argument)

# get withinsubject CIs
sfgend_wsci= Rmisc::summarySEwithin(data = highlowaov, measurevar = "dprime",
                       withinvars = "condition1", betweenvars = "testexpect", idvar = "participant_private_id")

## Automatically converting the following non-factors to factors: testexpect, condition1

#plot

fig1a <- ggplot(highlowaov,aes(x=condition1,y=dprime,fill=condition1))+ facet_grid(~testexpect) + 
  #geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = condition1, y = dprime),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
    geom_line(data=sfgend_wsci,aes(y=dprime, group=1), size=1)+ 
  #stat_summary(fun="mean", geom="point", colour="darkred", size=3) + 
    geom_pointrange(data=sfgend_wsci, aes(y=dprime, ymin=dprime, ymax=dprime), size=.8, color="darkred")+
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "Sensitivity(d')", x = "Typeface") + theme(legend.position = "none") + 
    geom_label_repel(data=sfgend_wsci, aes(y=dprime, label=round(dprime, 2)), min.segment.length = 0, seed = 42, box.padding = 0.5) +
  theme_cowplot(font_size = 14) + 
 theme(axis.title=bold, legend.position = "none")
 
fig1a_diff <- ggplot(highlowaov_wide,aes(x=testexpect,y=Difference, fill=testexpect)) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .18),size = 1, alpha = 0.2) +
  geom_boxplot(aes(x = testexpect, y = Difference),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
  stat_summary(fun.data="mean_cl_boot", colour="darkred", size=.8)+
  #geom_line(data=sfarial_wsci,aes(y=mean_acc, group=1), size=1)+ 
  #geom_pointrange(data=sfarial_wsci, aes(y=mean_acc, ymin=mean_acc-ci, ymax=mean_acc+ci), size=.5, color="red")+ 
  scale_colour_brewer(palette = "Accent")+
  scale_fill_brewer(palette = "Accent") +
  labs(y = "Test Difference (Sans Forgetica - Arial", x = "Test Expectancy")+
     theme_cowplot(font_size=14)+ 
  theme(legend.position = "none") +
  theme(axis.title =bold) + 
  geom_hline(yintercept = 0, linetype="dotted") + 
   geom_label_repel(data=highlowwide_mean, aes(y=mean, label=round(mean, 2)), seed=42, box.padding=0.8)

Analysis

#ANOVA


a1 <- aov_ez("participant_private_id", "dprime", highlowaov, 
             between = c("testexpect"), within=c("condition1")) # mixed

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                        Sum Sq num Df Error SS den Df  F value    Pr(>F)    
## (Intercept)           296.652      1  166.184    229 408.7834 < 2.2e-16 ***
## testexpect              2.980      1  166.184    229   4.1058  0.043896 *  
## condition1              1.818      1   38.786    229  10.7344  0.001215 ** 
## testexpect:condition1   0.735      1   38.786    229   4.3369  0.038405 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

#kable(summary(a1))

JOLs

Read in JOL data

# get JOls from raw data

high

jol_high<- highdata %>% 
  mutate(testexpect="high")
low

jol_low<-datasetlow %>%
  mutate(testexpect="low")
jol_high_low <- rbind(jol_high, jol_low)

#bind high and low
jols<-jol_high_low %>% janitor::clean_names(.) %>% dplyr::mutate(date=as.Date(utc_date)) %>%  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020"|date=="0010/07/2020" | date=="09/06/2020",  zone_type=="response_slider_endValue" | zone_type=="response_text_entry")


jols$response<-as.numeric(jols$response)


jols1<- jols %>%
  dplyr::select(participant_private_id, response, testexpect) %>%
  dplyr::mutate(cond=rep(1:2, 231), font=ifelse(cond==1, "SF", "A")) %>%
  tidyr::drop_na() %>% 
  dplyr::mutate(testexpect=ifelse(testexpect=="low", "Low Test Expectancy", "High Test Expectancy"), font=ifelse(font=="A", "Arial", "Sans Forgetica"))

Plot JOls

# plot JOLs


# get withinsubject CIs
sfgenjol_wsci= Rmisc::summarySEwithin(data = jols1, measurevar = "response",
                       withinvars = "font", betweenvars = "testexpect", idvar = "participant_private_id")

## Automatically converting the following non-factors to factors: testexpect, font

jols1mean <- jols1 %>%
  dplyr::group_by(testexpect, font) %>%
  dplyr::summarise(mean1=mean(response))

## `summarise()` regrouping output by 'testexpect' (override with `.groups` argument)

JOL_wide<- jols1 %>% 
  dplyr::select(participant_private_id, response, testexpect, font) %>%
  tidyr::pivot_wider(names_from = "font", values_from = "response") %>%
  dplyr::mutate(Difference=`Sans Forgetica` - Arial)

JOL_mean <- JOL_wide %>%
  dplyr::group_by(testexpect) %>%
  dplyr::summarise(mean=mean(Difference, na.rm=TRUE))

## `summarise()` ungrouping output (override with `.groups` argument)

figjol <- ggplot(jols1,aes(x=font,y=response,fill=font))+ facet_grid(~testexpect) + 
  #geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = font, y = response),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
      geom_line(data=sfgenjol_wsci,aes(y=response, group=1), size=1)+ 

    #stat_summary(fun="mean", geom="point", colour="darkred", size=3)+
  geom_pointrange(data=sfgenjol_wsci, aes(y=response, ymin=response, ymax=response), size=.8, color="darkred")+
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "Judgements of Learning", x = "Typeface") + theme(legend.position = "none") +
   geom_label_repel(data=sfgenjol_wsci, aes(y=response, label=round(response, 2)), seed = 42, box.padding = 0.8)+
  theme_cowplot() +
theme(axis.title=bold, legend.position = "none")


fig2b_diff <- ggplot(JOL_wide,aes(x=testexpect,y=Difference,fill=testexpect)) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) +
  geom_boxplot(aes(x = testexpect, y = Difference),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
  stat_summary(fun.data="mean_cl_boot", colour="darkred", size=.8)+
   #stat_summary(fun="mean", geom="point", colour="darkred", size=3)+
 # geom_line(data=sfgenjol_wsci,aes(y=jols, group=1), size=1)+ 
 # geom_pointrange(data=sfgenjol_wsci, aes(y=jols, ymin=jols-ci, ymax=jols+ci), size=.3, color="red")+ 
  scale_colour_brewer(palette = "Accent")+
  scale_fill_brewer(palette = "Accent") +
  labs(y = "JOL Difference (Sans Forgetica - Arial)", x = "Test Expectancy") +  theme(legend.position = "none")+ 
  geom_label_repel(data=JOL_mean, aes(y=mean , label=round(mean, 2)), seed = 42, box.padding = 0.8) + 
   theme_cowplot(font_size=14)+ 
  theme(legend.position = "none") +
  geom_hline(yintercept = 0, linetype="dotted") + 
  theme(axis.title = bold)



figjol

#6.67

Analysis

#anova JOLs
a1 <- aov_ez("participant_private_id", "response", jols1, 
             between = c("testexpect"), within=c("font")) # mixed

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                  Sum Sq num Df Error SS den Df  F value    Pr(>F)    
## (Intercept)     1575566      1   403702    221 862.5179 < 2.2e-16 ***
## testexpect        29254      1   403702    221  16.0145 8.579e-05 ***
## font               1803      1    14729    221  27.0520 4.509e-07 ***
## testexpect:font       9      1    14729    221   0.1334    0.7152    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

RTs

Read in RT data

#raw gorilla data and extract RTs

datasetlow$testexpt<-"low"
highdata$testexpt<-"high"

rt_high_low <- rbind(datasetlow, highdata)

rt<-rt_high_low %>% janitor::clean_names(.) %>% mutate(date=as.Date(utc_date)) %>%  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020"|date=="0010/07/2020" | date=="09/06/2020", zone_type=="continue_button", display=="study") 

# get RT and make numeric (Gorilla does not do this)
rt$reaction_time<-as.numeric(rt$reaction_time)

rt1<- rt %>% 
  dplyr::group_by(participant_private_id, condition, testexpt) %>% 
  dplyr::select(participant_private_id, condition, testexpt, reaction_time) %>%
  dplyr::mutate(sdabove = mean(reaction_time, na.rm=TRUE) +  2.5*sd(reaction_time, na.rm=TRUE)) %>%
    dplyr::filter(reaction_time > 150, reaction_time < sdabove) %>%
  dplyr::summarise(mean_rt= mean(log(reaction_time))) %>%
   mutate(testexpt=ifelse(testexpt=="low", "Low Test Expectancy", "High Test Expectancy"), font=ifelse(condition=="normal", "Arial", "Sans Forgetica")) %>%
  select(-condition) %>%
  ungroup()

Plot RT data

rt_wide <- rt1 %>%
  dplyr::select(-condition)%>%
  tidyr::pivot_wider(names_from="font", values_from = "mean_rt")%>%
  dplyr::mutate(Difference=`Sans Forgetica` - Arial)


rt_wide_mean <- rt_wide %>% 
  dplyr::group_by(testexpt) %>%
  dplyr::summarise(mean=mean(Difference))

## `summarise()` ungrouping output (override with `.groups` argument)

# get withinsubject CIs
sfgenrt_wsci= Rmisc::summarySEwithin(data = rt1, measurevar = "mean_rt",
                       withinvars = "font", betweenvars = "testexpt", idvar = "participant_private_id")

## Automatically converting the following non-factors to factors: testexpt, font

#plot fig

figrt <- ggplot(rt1,aes(x=font,y=mean_rt,fill=font))+ facet_grid(~testexpt) + 
  #geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = font, y = mean_rt),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
        geom_line(data=sfgenrt_wsci,aes(y=mean_rt, group=1), size=1)+ 

   geom_pointrange(data=sfgenrt_wsci, aes(y=mean_rt, ymin=mean_rt, ymax=mean_rt), size=.8, color="darkred")+
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "log(Study Times)", x = "Typeface") + theme(legend.position = "none")+
   geom_label_repel(data=sfgenrt_wsci, aes(y=mean_rt, label=round(mean_rt, 2)), min.segment.length = 0, seed = 42, box.padding = 0.8) + 
  theme_cowplot(font_size=14) + 
  theme(legend.position = "none", axis.title = bold)



figrt_diff <- ggplot(rt_wide,aes(x=testexpt,y=Difference,fill=testexpt)) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) +
  geom_boxplot(aes(x = testexpt , y = Difference),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
   stat_summary(fun.data="mean_cl_boot", colour="darkred", size=.8)+
   #stat_summary(fun="mean", geom="point", colour="darkred", size=3)+
  #geom_line(data=sfgenrt_wsci,aes(y=mean_rt, group=1), size=1)+ 
  #geom_pointrange(data=sfgenrt_wsci, aes(y=mean_rt, ymin=mean_rt-ci, ymax=mean_rt+ci),size=.3, color="red") + 
  scale_colour_brewer(palette = "Accent")+
  scale_fill_brewer(palette = "Accent") +
  labs(y = "Time Difference (Sans Forgetica - Arial)", x = "Test Expectancy") + theme(legend.position = "none") + 
   geom_label_repel(data=rt_wide_mean, aes(y=mean, label=round(mean, 2)), seed = 42, box.padding = 0.5) + 
theme_cowplot(font_size=14) + 
    geom_hline(yintercept = 0, linetype="dotted") + 
  theme(legend.position = "none", axis.title = bold)



ggsave("figrt.png", width=8, height=4, dpi=300)

figrt

figrt_diff

#write.csv(rt2, file="rt_high_low.csv")

#ttestBF(x=rt2$normal, y=rt

Analysis

#anova RTs

a1 <- aov_ez("participant_private_id", "mean_rt", rt1, 
             between = c("testexpt"), within=c("condition")) # 

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                     Sum Sq num Df Error SS den Df    F value    Pr(>F)    
## (Intercept)        20706.2      1  168.431    229 28152.2648 < 2.2e-16 ***
## testexpt               1.1      1  168.431    229     1.5354    0.2166    
## condition              0.3      1    1.797    229    33.0251 2.884e-08 ***
## testexpt:condition     0.0      1    1.797    229     1.1292    0.2891    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

#kable(summary(a1))

Combine Plots

fig1_plot <- plot_grid(
  fig1a, figjol, figrt,
  labels = "AUTO", ncol= 1, nrow = 3
)

ggsave("figexpt1.png", width=10, height=14, dpi=500)

fig1_diff <- plot_grid(fig1a_diff, fig2b_diff, figrt_diff, labels = "AUTO", ncol=, nrow=3)

ggsave("figexpt1_diff.png", width=10, height=14, dpi=500) 

fig1_diff_all <- plot_grid(fig1a, fig1a_diff, figjol, fig2b_diff,figrt, figrt_diff, labels = "AUTO", ncol=2, nrow=3)

ggsave("figexpt1_diff_all.png", width=12, height=14, dpi=500)